HIV exploits host cells via an intricate network of interactions. The HIV-1 Human Protein Interaction Database (a public resource available at the NCBI website) captures this information from the published literature. All published interactions involving HIV-1 and human proteins have been catalogued. Here we characterise the general properties of the HIV interacting host proteins and the nature of the HIV-host perturbation. In particular, we investigate (i) whether HIV preferentially interacts with key (highly connected and highly central) proteins, (ii) known phenotypic properties of host proteins inferred from essentiality and disease-association data, and (iii) molecular function of the host proteins (using the Gene Ontology, GO) to identify those host processes most strongly associated with specific HIV interactions. We demonstrate a significantly greater propensity for HIV to interact with highly connected and highly central host proteins. Unexpectedly, we find there are no general associations between HIV interaction and inferred essentiality, despite highly connected and highly central proteins having a tendency to be lethal mouse knock-outs. Similarly, we find a tendency for HIV to avoid interactions with proteins linked to non-lethal deleterious mutations as inferred by disease-associated genes. This suggests that the perturbation of the host system by retroviruses has evolved to use key proteins and there is a selection pressure to minimise interactions with 'essential' host proteins. We hypothesise HIV must exploit specific host proteins to achieve its biology, irrespective of their connectivity and centrality. Focussing on detailed biological functions (from GO) demonstrates the directionality and complexity of both pro-host (promoting HIV's replication cycle) and pro-pathogen (the host response to infection) interactions with specific cellular functions. Multiple functions, e.g. the immune response and apoptosis, are targeted by multiple viral proteins, being both up- and down-regulated. Crucially, we find that these functional categories are enriched for highly connected and highly central proteins in the host. Thus, HIV's propensity to interact with highly connected and highly central proteins is most probably a consequence of interactions with particular cellular functions rather than related to global network properties. Thus, the specificity of HIV-host interactions is intricate and not fully explained by the network. This detailed mapping demonstrates that HIV biology is involved in a complex 'arms race' and highlights the ancient association of retroviruses with their hosts.

HIV exploits the cell via an intricate network of interactions. Here we characterise the general properties of HIV interacting proteins and the nature of the HIV-host perturbation. We demonstrate a greater propensity for HIV to interact with highly connected and central host proteins. Unexpectedly, we find there are no associations between HIV interaction and inferred essentiality, despite highly connected proteins having a tendency to be lethal knock-outs. This suggests that the perturbation of the host system by retroviruses has evolved to use key proteins and there is a selection pressure to minimise interactions with 'essential' host proteins. We hypothesise HIV must exploit specific host proteins to achieve its biology, irrespective of their connectivity and centrality. Focussing on detailed biological functions demonstrates multiple functions are targeted by multiple viral proteins. Crucially, we find that functional categories are enriched for highly connected and central proteins in the host. Thus, HIV's propensity to interact with highly connected and central proteins is most probably a consequence of interactions with particular cellular functions rather than related to global network properties. This detailed mapping demonstrates that HIV biology is involved in a complex 'arms race' and highlights the ancient association of retroviruses with their hosts.

Network biology is concerned with representing cellular interactions and associated biological systems as networks. Substantial amounts of interaction information are curated from the scientific literature. The interests and expertise of researchers, their peers and collaborators, funding streams, public interest and nascent protocols and equipment all guide contemporary research. As such, there is undoubtedly bias in scientific research and hence the resulting literature. Robust analyses are reliant upon accurate data, for instance, in calculating the degree of a particular protein in a protein-interaction network. However, it is feasible that this protein, and hence its interactions, are affected by ascertainment bias. Specifically, if highly studied, it is likely that more interactions involving the protein will be known, thus distorting the observed network structure. We have previously evaluated node-level bias based on the proteins' publication counts in PubMed. Here we expand this work and present a novel method to evaluate ascertainment bias on interaction networks as a whole. We use rejection sampling to generate randomised sets of protein-protein interactions with the same pattern of bias as the literature-curated yeast interaction network. Applying this method to evaluate classic network properties, we find that rejection sampling performs favourably compared to na�ve random sampling. Ultimately, this method will not only greatly facilitate the assessment of bias in literature-based interactomes, it will also enable a more rigorous insight into the significance of biological network properties.

HIV's exploitation of the human host for replication involves highly specific interactions with the cellular system. The persistence of the virus is dependent on an intricate network of biological interactions. There is presently extensive knowledge of molecular interactions within the scientific literature. However, manual curation of these would be an enormously time-consuming task. Here we present a procedure for automated host-pathogen focused data extraction. We focus on mining HIV-1-host protein interactions, using the gold-standard HIV-1 Human Protein Interaction Database, containing 5,127 interactions manually extracted from 14,312 references, as a benchmark for evaluation. Our process involves employing a combination of named entity recognisers (such as BANNER) and the highest performing event extraction tools (e.g. from the BioNLP 2009 shared task). We replicated the original dataset to an acceptable degree, however with compromises devised to conform to the design of tools presently available. Notably, event extraction tools are limited to obtaining interactions from 9 event types, which are not compatible with all of the interactions in HIV biology. Thus, we have developed an appropriate ontology to encompass the full spectrum of host-pathogen interactions and demonstrate improved coverage. As a result, we demonstrate the effective use of sophisticated text mining tools to functionally recreate a large database of host-pathogen based interactions. This provides a platform for full-scale automated extraction of host-pathogen interactions from all the available literature (20 million+ online references). Not only will such an approach improve our understanding of host-pathogen biology, it may also permit informed development of pharmaceutical interventions.

Over the last ten years, the study of the properties of protein interaction networks (PINs) has become a major field of computational biology, having impact on a wide range of important biological questions, most notably in regard to the function and evolution of systems. Although deficiencies in the data are often recognized, adequate controls for bias in both high-throughput and literature-curated data are usually absent, rendering many studies of the topological properties of PINs inconclusive. Both noise and incompleteness are known to influence network properties, and sampling schemes capable of incorporating additional information have been proposed. However, null models to deal appropriately with the effects of ascertainment bias on network data are not currently available. Here we present a method to correct for ascertainment bias in literature-curated protein interaction networks, incorporating a statistical model of an idealized gene selection/ experiment system. The performance of the model is evaluated on synthetic data sets, and contrasted with the results obtained using a naive uniform random selection null model. Under the bemodel, well-studied proteins are reported to be associated with significantly higher degree, even when such an association is not present in the real network. However, when ascertainment bias is estimated by fitting the observed data to our model, a more appropriate hypothesis test can be implemented to correctly assess whether such an association is really present. With the growth of text mining as a source of primary data for applications.

Substantial amounts of information are curated from scientific literature. However, there is undoubtedly bias in scientific research and hence the resulting literature. analyses are reliant upon accurate data to draw conclusions; for instance, calculating the degree of a particular protein in a protein-interaction network. However, it is feasible that this protein, and hence its interactions, are affected by bias. Specifically, if highly studied, it is likely that more interactions will be known, distorting the observed network structure. To compensate, we devised a novel method to evaluate ascertainment bias based on a gene's publication count in PubMed. If a gene has a high publication count, it is inferred that it is highly studied. We used rejection sampling to generate sets of protein-protein interactions with the same patterns of bias. Applying this method to test hypotheses we find that rejection sampling performs favourably compared to randomly sampling directly from the unweighted distributions.

Scientific research, like many other social activities, is subject to trends. Certain research topics will become more attractive, exciting or fundable as public interests change and new experimental techniques are developed. Here we use GLASS (Gene LAyout by Semantic Similarity), a new methodology for data visualization in a whole-genome context, to explore research trends in yeast biology. Following the public release of the Saccharomyces cerevisiae genome in 1996, a rapid escalation of the number of publications per year is seen for many genes. However, this expansion is not uniform over the whole genome, revealing differences in the levels of scientific activity associated with different aspects of yeast biology.

The completion of the human genome sequence is permitting a detailed characterisation of the genetics underpinning human disease. Whilst we cannot date the first emergence of a disease mutation, we can date the emergence of the linked disease gene using common ancestors from the tree of life. We find that progenitors of all genes have a tendency to be ancient and that there is a distinct peak of disease after vertebrate divergence. As functional evolution is driven by gene duplication, we must consider duplicates when dating disease gene emergence. Interestingly, the proportion of disease-associated genes arising from duplicates in each time period are relatively constant over evolutionary time. For instance, whilst the majority of disease-associated genes emerged in the clade Chordata-Euteleostomi the majority of genes also evolved in this period. This would suggest that the tolerance of disease-associated mutations has also remained constant. We also investigated the proportion of disease-associated genes in each time period for singletons and found a steady decline over time. This suggests that in this case the tolerance of disease-associated mutations has decreased with increasing functional complexity. In conclusion, the emergence of disease-associated genes has been on-going throughout the evolution of life at a relatively constant rate for duplicates, while for singletons their association with disease has declined. We suggest human disease is a by-product inherent to biological systems dependent on mutation for functional innovation.

The recently published HIV-1, Human Protein Interaction Database has provided the HIV-research community with an extensive set of protein-protein interactions (PPIs). In this work, we use a biclustering to identify clusters of human proteins in the HIV-human PPI network. We identify 272 statistically significant human protein clusters. We show that many of these clusters consist of biologically related proteins that respond in similar ways during the course of HIV-1 infection. In addition, we show that biclustering is a useful method for analaysing PPI networks.

HIV exploits the cell via an intricate network of interactions. Here we characterise the general properties of HIV interacting proteins and the nature of the HIV-host perturbation. We demonstrate a greater propensity for HIV to interact with highly connected and central host proteins. Unexpectedly, we find there are no associations between HIV interaction and inferred essentiality, despite highly connected proteins having a tendency to be lethal mouse knock-outs. This suggests that the perturbation of the host system by retroviruses has evolved to use key host proteins. We hypothesise HIV must exploit specific these host proteins to achieve its biology, irrespective of their connectivity and centrality. Focussing on Gene Ontology annotations demonstrates multiple biological functions are targeted by multiple viral proteins. Crucially, we find that these functional categories are enriched for highly connected and central proteins in the host. Thus, HIV's propensity to interact with highly connected and central proteins is most probably a consequence of interactions with particular cellular functions rather than related to global network properties. This detailed mapping demonstrates that HIV biology is involved in a complex 'arms race' and highlights the ancient association of retroviruses with their hosts.

The recently published HIV-1, Human Protein Interaction Database has provided the HIV-research community with an extensive set of protein-protein interactions (PPIs). In this work, we use a biclustering to identify clusters of human proteins in the HIV-human PPI network. We identify 272 statistically significant human protein clusters. We show that many of these clusters consist of biologically related proteins that respond in similar ways during the course of HIV-1 infection. In addition, we show that biclustering is a useful method for analaysing PPI networks.

HIVs use of the host cell for replication involves highly specific interactions with the cellular system. The viral replication cycle is thus dependent on an intricate network of protein interactions, both between the virus and the host, and the various host proteins that are subverted. Here we investigate the general properties of these HIV interacting host proteins and the nature of the HIV-host perturbation. To do this we exploit the HIV-1 Human Protein Interaction Database available at NCBIs website. This interaction data can be organised as a protein interaction network and properties of interacting human proteins explored. We first investigate the topological properties of the host network and find that HIV has a propensity to interact preferentially with hubs and bottlenecks that respectively have high connectivity and betweenness centrality. While such topological features of the human network are essential, that is they have a tendency to be lethal mouse knock-outs, we find there is no relationship between HIV interaction and inferred essentiality. In support of this, we also observe a tendency for HIV to avoid interacting with non-lethal deleterious mutations as inferred by disease-associated genes. This suggests that the perturbation of the host system by retroviruses has evolved to use key proteins, there is a selection pressure to minimise interactions with 'essential' host proteins. We hypothesise HIV must exploit specific host proteins to achieve its biology, irrespective of their connectivity and centrality. To investigate this, we explored the specific types of functional categories that HIV perturbs. Crucially, we find that these functional categories are enriched for hubs and bottlenecks in the host. Thus, HIV's propensity to interact with hubs and bottlenecks is most probably a consequence of interactions with particular cellular functions rather than related to global network properties. In conclusion, the specificity of HIV-host interactions is intricate and not fully explained by the topological features of the interaction network. This detailed mapping highlights the ancient association of retroviruses with their hosts.

The recently published HIV-1, Human Protein Interaction Database has provided the HIV-research community with an extensive set of protein-protein interactions (PPIs). In this work, we use a biclustering to identify clusters of human proteins in the HIV-human PPI network. We identify 272 statistically significant human protein clusters. We show that many of these clusters consist of biologically related proteins that respond in similar ways during the course of HIV-1 infection. In addition, we show that biclustering is a useful method for analaysing PPI networks.

HIV's use of the host cell for replication involves highly specific interactions with the cellular system. The viral replication cycle is thus dependent on an intricate network of protein interactions between: (i) the viral proteins themselves, (ii) the virus and the host, and (iii) the various host proteins that are subverted. Here we investigate the general properties of HIV interacting host proteins and the nature of the HIV-host perturbation. To do this we exploit the HIV-1, Human Protein Interaction Database available at NCBI's website. This invaluable resource currently comprises 1,435 human genes encoding 1,448 proteins that interact with HIV-1 by 2,589 unique interactions. This interaction data can be organised as a protein interaction network and properties of interacting human proteins explored. In particular, we investigate (i) whether HIV preferentially interacts with key proteins (highly connected and bottlenecks), (ii) known phenotypic properties of host proteins inferred from essentiality data, and (iii) molecular function of the host proteins (using the Gene Ontology, GO) to identify those host processes most strongly associated with specific HIV interactions. We demonstrate a significantly greater propensity for HIV to interact with highly connected and bottleneck host proteins. We find there are no general associations between HIV interaction and essentiality. Investigating in detail the specific types of functional categories that HIV interacts with by gene and interaction type, we reveal the highly detailed and complex nature of HIV's infection of the host. For example, certain functional categories, e.g. the immune response and apoptosis, are targeted by multiple viral proteins, being both up- and down-regulated. Focussing on detailed biological functions (from GO) demonstrates the directionality and complexity of both pro-host (promoting HIV's replication cycle) and pro-pathogen (the host response to infection) interactions with specific cellular functions. Thus, the specificity of HIV-host interactions is intricate and not fully explained by the topological features of the network. This detailed mapping demonstrates that HIV biology is involved in a complex 'arms race' and highlights the ancient association of retroviruses with their hosts.

HIV exploits host cells via an intricate network of interactions. The HIV-1 Human Protein Interaction Database (a public resource available at the NCBI website) captures this information from the published literature. All published interactions involving HIV-1 and human proteins have been catalogued. Here we present an interactive visualisation of the HIV-human protein interaction network, forming the basis of a detailed map for tracking both direct and indirect molecular interactions involved in HIV-1 replication and pathogenesis. Incorporation of additional information, for example, expression data, cell-type and known drug targets, helps to place these interactions into their biological context. Ultimately this approach has the potential to accelerate the development of more effective and sophisticated antiviral and vaccine interventions. This integrative methodology mirrors the shift in molecular biology from focussing on a small set of components to having an understanding and appreciation of an entire system.